openai/openai-realtime-agents
OfficialThis is a simple demonstration of more advanced, agentic patterns built on top of the Realtime API.
OpenAI Realtime Agents is a TypeScript demonstration repository from OpenAI that showcases advanced agentic patterns for voice-enabled AI using the Realtime API and Agents SDK Clone the repo and adapt the voice agent patterns to build real-time conversational AI with WebSockets and the OpenAI Age...
Best for
Our Review
OpenAI Realtime Agents is the official OpenAI demo for realtime voice agents -- 6,818 GitHub stars as of April 2026. Built in TypeScript and released under the MIT license, it provides production-ready patterns using the Realtime API over WebSockets plus the OpenAI Agents SDK.
What OpenAI Realtime Agents does:
- •Realtime voice conversations stream audio input and output for natural, low-latency spoken interactions with AI.
- •Agent state management track conversation history, tools, and orchestration through the Agents SDK.
- •Tool calling integration connect voice sessions to external APIs and functions during live calls.
- •Multi-agent support run multiple specialized agents in one session with handoffs and supervision.
- •TypeScript reference code copy structured modules for chat agents, supervisors, and configs into your app.
Patterns demonstrated:
- •Chat-Supervisor realtime voice agent handles basics while a text model like gpt-4.1 manages tools and complex logic.
- •Sequential Handoff pass users between domain-specific voice agents to boost accuracy in customer service flows.
Ecosystem fit:
- •OpenAI Agents SDK unifies behaviors, events, and realtime streaming in one toolkit.
- •Next.js app structure deploy as a web server with browser-based voice demos out of the box.
Getting started:
Clone the repo and run npm i to install dependencies. Add your OPENAI_API_KEY to .env. Start with npm run dev and visit localhost:3000. Switch patterns via the Scenario dropdown. Adapt src/app/agentConfigs for custom agents. Full setup takes under 5 minutes for TypeScript devs.
Limitations:
No npm package or releases -- clone the repo and build your own version. Requires paid OpenAI API key -- free tier limits realtime usage. Next.js focused -- adapt for other frameworks like Express or Deno.
Cons
- No npm package or releases -- clone the repo and build your own version.
- Requires paid OpenAI API key -- free tier limits realtime usage.
- Next.js focused -- adapt for other frameworks like Express or Deno.
- Demo scale only -- add persistence and auth for production traffic.
Our Verdict
Developers building voice AI pick OpenAI Realtime Agents for official patterns that integrate Realtime API with Agents SDK. You save weeks on architecture by cloning chat-supervisor or handoff flows. Stars hit 6,818 for a reason -- it shows real agentic voice in TypeScript.
Vibe builders add voice to apps fast with this demo. Run it locally to test conversational AI before scaling. Focus on your product while using proven OpenAI code.
Skip OpenAI Realtime Agents if you avoid OpenAI costs or need multi-provider voice like LiveKit. Choose it when tight Realtime API integration drives your stack.
Frequently Asked Questions
What is OpenAI Realtime Agents and what does it demonstrate?
OpenAI Realtime Agents is an official demo repo from OpenAI. It shows advanced voice agent patterns with the Realtime API and Agents SDK. Key demos include chat-supervisor for tool delegation and sequential handoff for specialist agents.
Is OpenAI Realtime Agents free and open source?
OpenAI Realtime Agents is fully open source under the MIT license. You clone and modify the TypeScript code at no cost. Runtime uses paid OpenAI API calls beyond free tier limits.
OpenAI Realtime API vs WebSocket API -- what is the difference?
The OpenAI Realtime API is a WebSocket-based protocol for voice streaming. It handles audio I/O, events, and tools natively. Plain WebSockets require manual audio processing -- Realtime API simplifies agentic voice.
How do I build a voice agent using this repo?
Clone OpenAI Realtime Agents and customize agentConfigs in src/app. Define behaviors with OpenAI Agents SDK for tools and state. Run `npm run dev` to test voice in browser. Deploy to Vercel for production in 2026.
What prerequisites are needed to run OpenAI Realtime Agents?
You need Node.js, npm, and an OpenAI API key. Install with `npm i` in the Next.js project. No GPU required -- runs on CPU for demos. Browser mic access enables voice input.
What is openai-realtime-agents?
OpenAI Realtime Agents is a TypeScript demonstration repository from OpenAI that showcases advanced agentic patterns for voice-enabled AI using the Realtime API and Agents SDK Clone the repo and adapt the voice agent patterns to build real-time conversational AI with WebSockets and the OpenAI Age...
What license does openai-realtime-agents use?
openai-realtime-agents uses the MIT license.
What are alternatives to openai-realtime-agents?
Search My AI Guide for similar tools in this category.
Great for: Pro Vibe Builders
Skip if: You need something more beginner-friendly or guided
Open source & community-verified
MIT licensed — free to use in any project, no strings attached. 6,826 developers have starred this, meaning the community has reviewed and trusted it.
Reviewed by My AI Guide for relevance, quality, and active maintenance before listing.